Introduction

Pacific giant salamanders (Dicamptodon tenebrosus) is the largest terrestrial salamander in the Pacific Northwest (Pentranka 1998). Though the species is listed as Least Concern on IUCN Red List (Baillie et al. 2004), it is considered endangered in British Columbia, Canada where it has limited range (Munro 1993). Logging is the primary threat to the salamanders especially for larvae where the sapcies relied strongly on the stream for living. Research found larvae to be about four times more abundant in streams passing through old growth forest section than clear cut ones (Corn and Bury 1989). This decline is because streams impacted by fine sediments can lead to siltation, a result of logging. Increased siltation can fill rocky interstices under cover objects, where salamanders usually live and hide from predators, thus greatly dropped the population in logged areas (Welsh and Ollivier 1998). However, other studies also reported higher densities of the species in recently logged sites, which might be a result of increased primary productivity (Murphy et al. 1981). To further our understanding of logging impacts on Pacific giant salamanders, we analyzed observations in clear cut and old growth forest sections in Mack Creek to assess the influence on salamander population size, channel distribution, and weight.

A nice photo of Pacific Giant Salamander

A nice photo of Pacific Giant Salamander

Photo credit to Will Flaxington. Downloaded from CalPhotos on December 3rd, 2019

Data and Methods

The monitoring data of Pacific giant salamanders was collected and provided by Andrews Forest LTER Site and made accessible by EDI Data Protal. In addition to Pacific giant salamander, the group also collected observations for cutthroat trout (Onchorhyncus clarkii) and Olympic torrent salamanders (Phyacotriton olympicus).

The data was collected from a section of clear cut forest and an upstream 500-year-old coniferous forest in Mack Creek (Gregory 2016). In this study, only data associated with Pacific giant salamander were used which narrowed the year range to 1993-2017. We summed and visualized the annual count of salamanders during this time period for clear cut (CC) and old growth (OG) forest sections. Then, we focused on observations in 2017 and added channel classification into analysis by selecting salamanders observed in cascades, pool, and side-channel. A chi-squared test was used to evaluate the associated effect of forest sections and channel classifications on salamander distribution. An unpaired two-sample t-test was used to test if the populations in CC and OG had equal means of weight in 2017. Lastly, we used one-way unbalanced ANOVA to compare the populations means across channel classification groups. For all tests, we checked assumptions if necessary and chose 95% significant level (\(\alpha\) = 0.05) for inference. All analyses and figures were prepared using R software version 3.6.1. Mack Creek Map Mack Creek map was made with Google Earth 2018 on December 3rd, 2019.

Results

A. Visually compare annual salamander counts in old growth (OG) and clear cut (CC) sections of Mack Creek.

The annual count of Pacific giant salamanders was visualized for comparison between two groups of forest sections (CC/OG) of Mack Creek from 1993-2017 (Figure 1). In both sections, the count of Pacific giant salamanders generally increased from 1993 to 2011, then fluctuated from 2011 to 2017 but did not drop back to the lowest levels in late 1990s. In most years, salamanders in OG section had greater counts and seemed to be more robust to survival pressures than those in CC section.

Figure 1: Annual count of Pacific giant salamanders in clear cut (CC) or old growth (OG) forest sections of Mack Creek (1993-2017). The red line and points represent the annual count of salamanders in CC section. The green line and points show the annual count of salamanders in OG section. Data source: Andrews Forest LTER Site.


B. Salamander counts by channel classification (cascades, pool, and side-channel) and forest sections of Mack Creek in 2017.

We summarized the number and proportion of Pacific giant salamanders observed in all channel and forest sections (Table 1). The porportions of salamaders spotted in CC and OG were similar for all three channel classifications, where most salamanders were found in channal cascades while least were found in pools.

Table 1: Numbers of salamanders within Mack Creek in 2017. Values indicated are counts and proportions (by sections) in channel units (cascades, pool, and side-channel). In CC section, 247 (67.1%) observations are from cascades, 31 (8.4%) from pools, and 90 (24.5%) from side channels. In OG section, 201 (62.8%) observations are from cascades, 45 (14.1%) from pools, and 74 (23.1%) from side channels. Data source: Andrews Forest LTER Site.

Sections Cascades Pool Side-channel
CC 247 (67.1%) 31 (8.4%) 90 (24.5%)
OG 201 (62.8%) 45 (14.1%) 74 (23.1%)

C. Is there a significant difference in where in the channel Pacific giant salamanders are located between the two forest condition (OG and CC)?

There is no significant effect of forest conditions (OG/CC) on where in the channel salamanders are located (cascades/pool/side-channel) (\(\chi\)2(2) = 5.54, p = 0.063). Salamanders from old growth and clear cut section in the forest have no different chances to be spotted in cascades (OG = 201 (62.8%), CC = 247 (67.1%)), pool (OG = 45 (14.1%), CC = 247 (67.1%)), and side-chanel (OG = 74 (23.1%), CC = 90 (24.5%)).


D. Compare weights of Pacific giant salamanders in OG and CC forest sections of the creek in 2017.

Although salamander weights in both forest sections (OG/CC) were positively skewed, the large data size ( CC = 368, OG = 320) allowed us to use a two-sample t-test to compare the population means in CC and OG (Figure 2).

Figure 2: The weight of Pacific giant salamanders in forest sections (CC/OG) in 2017. The distributions of weight in both forest sections were positively skewed. Salamanders in CC forest section have weight 7.78 \(\pm\) 9.9 gram (n = 368) while those in OG section have weight 6.7 \(\pm\) 9.04 gram (n = 320). Data source: Andrews Forest LTER Site.

For 2017 observations, the average weights of Pacific giant salamanders differed significantly for samples collected from OG section (6.7 \(\pm\) 9.04, n = 320) and CC section (7.78 \(\pm\) 9.9, n = 368) based on the two-sample t-test (t(683) = 1.49, p = 0.138, 95% CI = (-0.35, 2.49)).

Based on p-value, there was a 13.8% chance that we could have found two samples at least as different as ours if they were drawn from two populations with the same means, which is not signifcantly low.

In addition, if we took many samples and found the confidence intervals, we’d expect 95% of those confidence intervals to included the population mean differece as what we have obtained. Therefore, it’s highly likely that a calculated confidence interval contains the population mean difference, where 0 was included. The difference in salamaders weight between OG and CC forest sections was 1.08mm, with a negligible effect size (Cohen’s d = -0.11).

Therefore, we failed to reject the null hypothesis that the weight of Pacific giant salamanders measured in CC and OG forest sections were the same. The difference in forest conditions (CC/OG) did not impact salamanders’ weight in Mack Creak samples.


E. Weights of Pacific giant salamanders in pools, cascades, and side-channels of Mack Creek in 2017.

Figure 3: The weight of Pacific giant salamanders in different channel locations (cascades/pool/side-channel) in 2017. The distributions of weight in all three locations are positively skewed. Salamanders in cascades have weight 7.52 \(\pm\) 9.03 gram (n = 448, SE = 0.43), those in the pools have weight 9.3 \(\pm\) 13.62 gram (n = 76, SE = 1.56), and those in the side-channels have weight 5.68 \(\pm\) 8.27 gram (n = 164, SE = 0.65). Data source: Andrews Forest LTER Site.

Although the weight of Pacific giant salamanders is positively skewed, we can still conduct a one-way unbalanced ANOVA test because the variance are close (Levene’s F(2) = 2.39, p = 0.093) and sample sizes are large (cascades = 448, pool = 76, side-channel = 164) for the three groups.

The average weights of Pacific giant salamanders differed significantly across all channel classifications (one-way ANOVA with post-hoc Tukey’s HSD, F(2, 684) = 4.22, p = 0.015). This result means that salamanders spotted in at least one group of the cascades, pool, or side-channel had a different mean weight than the other groups. Pairwise comparison showed that mean weights of salamanders in side-channel and pool were significantly different (mean difference = -3.62, p adj = 0.017) while comparisons between pool and cascades (mean difference = 1.78, p adj = 0.286), side-channel and cascades (mean difference = -1.84, p adj = 0.084) did not show significant differences.

Finally, despite ANOVA assumptions were met and the test gave us effective and reasonable outputs by comparing means, a rank-based test compares the medians across the groups might make up the positive skewness of salamander weights.

Summary

References

Baillie, J., C. Hilton-Taylor, and S. N. Stuart. (2004). 2004 IUCN Red List of threatened species : a global species assessment.

Corn, P. S., and R. Bruce Bury. (1989). Logging in Western Oregon: Responses of headwater habitats and stream amphibians. Forest Ecology and Management 29:39–57.

Gregory S. V. (2016). Aquatic Vertebrate Population Study in Mack Creek, Andrews Experimental Forest, 1987 to present. Environmental Data Initiative. https://doi.org/10.6073/pasta/5de64af9c11579266ef20da2ff32f702 . Dataset accessed 11/19/2019.

Munro, W. T. (1993). Designation of endangered species, subspecies and populations by Cosewic.

Murphy, M. L., C. P. Hawkins, and N. H. Anderson. (1981). Effects of Canopy Modification and Accumulated Sediment on Stream Communities. Transactions of the American Fisheries Society 110:469–478.

Petranka, J. W. (1998). Salamanders of the United States and Canada. Smithsonian Institution Press, Washington and London.

Welsh, H., and L. Ollivier. (1998). Stream amphibians as indicators of ecosystem stress: a case study from California’s redwoods. Ecological Applications 8:1118–1132.